Data for: Systematic Review of Pedagogical Theory Operationalisation in Generative AI Educational Content Systems for Higher Education

This dataset contains the data extracted from the 23 primary studies included in a systematic review of how generative AI systems that produce educational content operationalise pedagogical theory in higher education. The studies were identified through a search of IEEE Xplore, Web of Science and Scopus for publications from 2020 to 2026, and selected through a procedure adapted from PRISMA. The search strings and selection steps are reported in the related article. Each row describes one study: its reference and DOI, year, educational setting, primary generative model, architecture, deployment pattern, type of content generated, the point at which pedagogical design enters the generation pipeline, and its primary and secondary evaluation paradigms. For the 21 higher-education studies, the dataset also gives the level at which each of five pedagogical theories is operationalised: Bloom's revised taxonomy, the Zone of Proximal Development and scaffolding, Cognitive Load Theory, Constructivism, and Self-Determination Theory. Levels use a four-level scale: 0, absent; 1, cited rhetorically; 2, formal design constraint; 3, empirically evaluated as a system outcome. This gives 105 paper-theory pairings. For each of these studies, the dataset also gives the highest level reached and a Pedagogical Alignment Score with uniform theory weights (the sum of the five levels divided by 15). The two K-12 studies were retained for the architectural analysis only, so their level, score and evaluation cells are empty. The data were extracted from the full text of each study with a structured extraction form. Levels were assigned on the basis of what each study reports, not what its system might possibly do; where the evidence was ambiguous between two levels, the lower level was assigned. The values match those reported in the related article. The codebook defines every column and code.